Anurag1734/cuda-error-resolution-analysis
07
1[2 {3 "post_stream": {4 "posts": [5 {6 "id": 192223,7 "name": "Michael Zhou",8 "username": "mizho",9 "avatar_template": "/user_avatar/discuss.pytorch.org/mizho/{size}/13049_2.png",10 "created_at": "2020-05-13T06:40:27.476Z",11 "cooked": "<p>I am having problems with a cuDNN RNN model I am trying to train on a set of natural language explanations and embeddings for the semantic parsing of texts. Here is what my RNN model architecture looks like on a simplified level:</p>\n<pre><code class=\"lang-auto\">class Cudnn_RNN:\n\n def __init__(self, num_layers, num_units, mode=\"lstm\", keep_prob=1.0, is_train=None, scope=\"cudnn_rnn\"):\n self.num_layers = num_layers\n self.rnns = []\n self.mode = mode\n if mode == \"gru\":\n rnn = tf.contrib.cudnn_rnn.CudnnGRU\n elif mode == \"lstm\":\n rnn = tf.contrib.cudnn_rnn.CudnnLSTM\n else:\n raise Exception(\"Unknown mode for rnn\")\n for layer in range(num_layers):\n rnn_fw = rnn(1, num_units)\n rnn_bw = rnn(1, num_units)\n self.rnns.append((rnn_fw, rnn_bw, ))\n\n def __call__(self, inputs, seq_len, keep_prob=1.0, is_train=None, concat_layers=True):\n outputs = [tf.transpose(inputs, [1, 0, 2])]\n for layer in range(self.num_layers):\n rnn_fw, rnn_bw = self.rnns[layer]\n output = dropout(outputs[-1], keep_prob=keep_prob, is_train=is_train)\n with tf.variable_scope(\"fw_{}\".format(layer)):\n out_fw, state_fw = rnn_fw(output)\n with tf.variable_scope(\"bw_{}\".format(layer)):\n inputs_bw = tf.reverse_sequence(output, seq_lengths=seq_len, seq_axis=0, batch_axis=1)\n out_bw, state_bw = rnn_bw(inputs_bw)\n out_bw = tf.reverse_sequence(out_bw, seq_lengths=seq_len, seq_axis=0, batch_axis=1)\n outputs.append(tf.concat([out_fw, out_bw], axis=2))\n if concat_layers is True:\n res = tf.concat(outputs[1:], axis=2)\n else:\n res = outputs[-1]\n res = tf.transpose(res, [1, 0, 2])\n state_fw = tf.squeeze(state_fw[0], [0])\n state_bw = tf.squeeze(state_bw[0], [0])\n state = tf.concat([state_fw, state_bw], axis=1)\n return res, state\n</code></pre>\n<p>The model is set up such that after data is loaded, it goes through pretraining, training, and then evaluation. For some reason the data is being loaded with no issues, but as soon as the model starts running it gets stuck, not even making it to the pretraining phase. I have included the data loading and model execution code below (the model fails to execute past the “sess.run(tf.global_variables_initializer())” line:</p>\n<pre><code class=\"lang-auto\">def pseudo_labeling(config, data):\n word2idx_dict, fixed_emb, traiable_emb, train_data, dev_data, test_data,pretrain_data,pretrain_data2 = data\n\n pretrain_test_data = (pretrain_data[0][:config.pretrain_test_size],pretrain_data[1][:config.pretrain_test_size],pretrain_data[2][:config.pretrain_test_size,:])\n pretrain_data = (pretrain_data[0][config.pretrain_test_size:config.pretrain_test_size+config.pretrain_train_size],pretrain_data[1][config.pretrain_test_size:config.pretrain_test_size+config.pretrain_train_size],pretrain_data[2][config.pretrain_test_size:config.pretrain_test_size+config.pretrain_train_size,:])\n\n lfs = get_lfs(config, word2idx_dict)\n identifier = \"_{}\".format(config.tag)\n\n with tf.variable_scope(\"models\", reuse=tf.AUTO_REUSE):\n regex = Pat_Match(config)\n match = Soft_Match(config,lfs['lfs'],np.array(lfs['rels'],np.float32),lfs['keywords'],lfs['keywords_rels'], lfs['raw_keywords'],mat=((fixed_emb, traiable_emb, )), word2idx_dict=word2idx_dict, pseudo=True)\n\n sess_config = tf.ConfigProto(allow_soft_placement=True)\n sess_config.gpu_options.allow_growth = True\n if os.path.exists('labeled_data.pkl'):\n with open('labeled_data.pkl', 'rb') as f:\n labeled_data = pickle.load(f)\n with open('unlabeled_data.pkl', 'rb') as f:\n unlabeled_data = pickle.load(f)\n with open('weights.pkl', 'rb') as f:\n lfs[\"weights\"] = pickle.load(f)\n else:\n with open('exp2pat.json','r') as f:\n exp2pat = json.load(f)\n exp2pat = {int(key):val for key,val in exp2pat.items()}\n lab_d = []\n unlab_d = []\n\n tacred_labeled = []\n tacred_unlabeled = []\n labeled_data = []\n unlabeled_data = []\n idxx = -1\n\n idx2rel = {val:key for key,val in constant.LABEL_TO_ID.items()}\n\n for x in tqdm(train_data):\n idxx+=1\n batch = [x[\"phrase\"]]\n res, pred = regex.match(batch)\n lfs[\"weights\"] += res[0]\n new_dict = {}\n if np.amax(res) > 0:\n\n x[\"rel\"] = pred.tolist()[0]\n x[\"logic_form\"] = np.argmax(res, axis=1).tolist()[0]\n new_dict['tokens'] = x['phrase'].token\n new_dict['start'] = min(x['phrase'].subj_posi,x['phrase'].obj_posi)+1\n new_dict['end'] = max(x['phrase'].subj_posi,x['phrase'].obj_posi)-1\n new_dict['rel'] = pred.tolist()[0]\n try:\n new_dict['pat'] = exp2pat[np.argmax(res, axis=1).tolist()[0]]\n lab_d.append(new_dict)\n except:\n new_dict['pat'] = -1\n unlab_d.append(new_dict)\n tacred_labeled.append((idxx,idx2rel[x['rel']]))\n labeled_data.append(x)\n else:\n tacred_unlabeled.append(idxx)\n new_dict['tokens'] = x['phrase'].token\n new_dict['start'] = min(x['phrase'].subj_posi,x['phrase'].obj_posi)+1\n new_dict['end'] = max(x['phrase'].subj_posi,x['phrase'].obj_posi)-1\n new_dict['rel'] = pred.tolist()[0]\n new_dict['pat']=-1\n x[\"rel\"] = 0\n unlab_d.append(new_dict)\n unlabeled_data.append(x)\n\n new_weight = np.array([elem for i, elem in enumerate(list(lfs['weights'])) if i in exp2pat],np.float32)\n new_weight = new_weight/np.sum(new_weight)\n lfs[\"weights\"] = lfs[\"weights\"] / np.sum(lfs[\"weights\"])\n\n with open('tacred_labeled.json','w') as f:\n json.dump(tacred_labeled,f)\n\n with open('tacred_unlabeled.json','w') as f:\n json.dump(tacred_unlabeled,f)\n\n with open('labeled_data.pkl','wb') as f:\n pickle.dump(labeled_data,f)\n with open('unlabeled_data.pkl','wb') as f:\n pickle.dump(unlabeled_data,f)\n with open('weights.pkl', 'wb') as f:\n pickle.dump(lfs[\"weights\"], f)\n\n with open('lab_d.pkl','wb') as f:\n pickle.dump(lab_d,f)\n with open('unlab_d.pkl','wb') as f:\n pickle.dump(unlab_d,f)\n with open('weights_d.pkl','wb') as f:\n pickle.dump(new_weight,f)\n\n random.shuffle(unlabeled_data)\n\n print('unlabdel data:',str(len(unlabeled_data)),'labeled data:',str(len(labeled_data)))\n\n dev_history, test_history = [], []\n dev_history2, test_history2 = [], []\n\n with tf.Session(config=sess_config) as sess:\n\n lr = float(config.init_lr)\n writer = tf.summary.FileWriter(config.log_dir + identifier)\n sess.run(tf.global_variables_initializer())\n\n print('---Pretrain-----')\n for epoch in range(config.pretrain_epoch):\n loss_list,pretrain_loss_lis,sim_loss_lis = [],[],[]\n for batch in get_pretrain_batch(config, pretrain_data, word2idx_dict):\n pretrain_loss_prt,sim_loss_prt,loss,_ = sess.run([match.pretrain_loss,match.sim_loss,match.pretrain_loss_v2,match.pre_train_op],feed_dict={match.pretrain_sents: batch['sents'], match.pretrain_pats: batch['pats'],match.pretrain_labels: batch['labels'],match.is_train:True})\n loss_list.append(loss)\n pretrain_loss_lis.append(pretrain_loss_prt)\n sim_loss_lis.append(sim_loss_prt)\n print(\"{} epoch:\".format(str(epoch)))\n print(\"loss:{} pretrain_loss:{} sim_loss:{}\".format(str(np.mean(loss_list)),str(np.mean(pretrain_loss_lis)),str(np.mean(sim_loss_lis))))\n pred_labels = []\n goldens = []\n prt_id = 0\n for batch in get_pretrain_batch(config,pretrain_data2,word2idx_dict,shuffle=False):\n prt_id+=1\n pp,ppp,pred_label = sess.run([match.prt_loss,match.prt_pred,match.pretrain_pred_labels],feed_dict={match.pretrain_sents: batch['sents'], match.pretrain_pats: batch['pats'],match.is_train:False,match.pretrain_labels: batch['labels']})\n pred_label = list(pred_label)\n golden = list(np.reshape(batch['labels'],[-1]))\n assert len(golden)==len(pred_label)\n pred_labels.extend(pred_label)\n goldens.extend(golden)\n p,r,f = f_score(pred_labels,goldens)\n print('PRF:',(p,r,f))\n if p>0.9 and r>0.9:\n break\n print('\\n')\n print('----Training----')\n for epoch in range(1, config.num_epoch + 1):\n pretrain_loss_lis,sim_loss_lis, labeled_loss_lis, unlabeled_loss_lis, hard_train_loss_lis, loss_lis = [],[],[],[],[],[]\n for batch1, batch2,batch3 in zip(get_batch(config, labeled_data, word2idx_dict), get_batch(config, unlabeled_data, word2idx_dict, pseudo=True),get_pretrain_batch(config,pretrain_data,word2idx_dict,pretrain=False)):\n batch = merge_batch(batch1, batch2)\n global_step = sess.run(match.global_step) + 1\n pretrain_loss,sim_loss, labeled_loss, unlabeled_loss, hard_train_loss,loss, _ = sess.run([match.pretrain_loss,match.sim_loss,match.labeled_loss,match.unlabeled_loss,match.hard_train_loss,match.loss, match.train_op], feed_dict=get_feeddict(match, batch,batch3))\n\n pretrain_loss_lis.append(pretrain_loss)\n sim_loss_lis.append(sim_loss)\n labeled_loss_lis.append(labeled_loss)\n unlabeled_loss_lis.append(unlabeled_loss)\n hard_train_loss_lis.append(hard_train_loss)\n loss_lis.append(loss)\n\n if global_step % config.period == 0:\n loss_sum = tf.Summary(value=[tf.Summary.Value(tag=\"model/loss\", simple_value=loss), ])\n writer.add_summary(loss_sum, global_step)\n writer.flush()\n\n (dev_acc, dev_rec, dev_f1), (dev_acc2, dev_rec2, dev_f12), (best_entro, best_bound), _ = log(config, dev_data,pretrain_data, word2idx_dict, match, sess, writer, \"dev\")\n (test_acc, test_rec, test_f1), (test_acc2, test_rec2, test_f12), _, _ = log(\n config, test_data,pretrain_data, word2idx_dict, match, sess, writer, \"test\", entropy=best_entro, bound=best_bound)\n writer.flush()\n\n print('\\n')\n print(\"{} epoch:\".format(str(epoch)))\n print(\"pretrain_loss:{} sim_loss:{} labeled_loss:{} unlabeled_loss:{} hard_train_loss:{} loss:{} best_bound:{}:\".format(str(np.mean(pretrain_loss_lis)),str(np.mean(sim_loss_lis)),str(np.mean(labeled_loss_lis)),str(np.mean(unlabeled_loss_lis)),str(np.mean(hard_train_loss_lis)),str(np.mean(loss_lis)),str(best_bound)))\n print(\"dev_acc:{} dev_rec:{} dev_f1:{} dev_acc_2:{} dev_rec_2:{} dev_f1_2:{}\\ntest_acc:{} test_rec:{} test_f1:{} test_acc_2:{} test_rec_2:{} test_f1_2:{}\".format(\n str(dev_acc),str(dev_rec),str(dev_f1),str(dev_acc2),str(dev_rec2),str(dev_f12),str(test_acc),str(test_rec),str(test_f1),str(test_acc2),str(test_rec2),str(test_f12)\n ))\n\n dev_history.append((dev_acc, dev_rec, dev_f1))\n test_history.append((test_acc, test_rec, test_f1))\n dev_history2.append((dev_acc2, dev_rec2, dev_f12))\n test_history2.append((test_acc2, test_rec2, test_f12))\n if len(dev_history) >= 1 and dev_f1 <= dev_history[-1][2]:\n lr *= config.lr_decay\n sess.run(tf.assign(match.lr, lr))\n\n max_idx = dev_history.index(max(dev_history, key=lambda x: x[2]))\n max_idx2 = dev_history2.index(max(dev_history2, key=lambda x: x[2]))\n max_acc, max_rec, max_f1 = test_history[max_idx]\n max_acc2, max_rec2, max_f12 = test_history2[max_idx2]\n print(\"acc: {}, rec: {}, f1: {}, acc2 {}, rec2 {}, f12 {}\".format(max_acc, max_rec, max_f1, max_acc2, max_rec2, max_f12))\n sys.stdout.flush()\n return max_acc, max_rec, max_f1, max_acc2, max_rec2, max_f12\n</code></pre>\n<p>Here are my system specifications:<br>\nTensorflow version: 1.14.0 (w/ GPU support)<br>\nOperating System: Linux 9.12<br>\nOS Distribution: Debian<br>\nOS Architecture: x86_64<br>\nPython version: 3.7.6<br>\nNLTK version: 3.4.5<br>\nCUDA version: 10.0<br>\ncuDNN version: 7.4.2<br>\nNVIDIA graphics card: Tesla T4<br>\nNVIDIA driver version: 410.104<br>\nCompiler version: GCC 6.3.0</p>\n<p>If anyone would like to share their thoughts on why my model is unable to properly execute to the pretraining phase and beyond, I would greatly appreciate it. Thank you.</p>",12 "post_number": 1,13 "post_type": 1,14 "posts_count": 2,15 "updated_at": "2020-05-13T06:52:04.108Z",16 "reply_count": 0,17 "reply_to_post_number": null,18 "quote_count": 0,19 "incoming_link_count": 29,20 "reads": 8,21 "readers_count": 7,22 "score": 146.6,23 "yours": false,24 "topic_id": 80974,25 "topic_slug": "model-execution-issues-with-cudnn-rnn",26 "display_username": "Michael Zhou",27 "primary_group_name": null,28 "flair_name": null,29 "flair_url": null,30 "flair_bg_color": null,31 "flair_color": null,32 "flair_group_id": null,33 "badges_granted": [],34 "version": 2,35 "can_edit": false,36 "can_delete": false,37 "can_recover": false,38 "can_see_hidden_post": false,39 "can_wiki": false,40 "read": true,41 "user_title": null,42 "bookmarked": false,43 "actions_summary": [],44 "moderator": false,45 "admin": false,46 "staff": false,47 "user_id": 31513,48 "hidden": false,49 "trust_level": 1,50 "deleted_at": null,51 "user_deleted": false,52 "edit_reason": null,53 "can_view_edit_history": true,54 "wiki": false,55 "post_url": "/t/model-execution-issues-with-cudnn-rnn/80974/1",56 "can_accept_answer": false,57 "can_unaccept_answer": false,58 "accepted_answer": false,59 "topic_accepted_answer": null,60 "can_vote": false61 },62 {63 "id": 192534,64 "name": "",65 "username": "ptrblck",66 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",67 "created_at": "2020-05-14T03:19:19.095Z",68 "cooked": "<p>It seems you are using TensorFlow, so I think you might get better support on StackOverflow, as the majority of the users here are more familiar with PyTorch. <img src=\"https://discuss.pytorch.org/images/emoji/apple/wink.png?v=9\" title=\":wink:\" class=\"emoji\" alt=\":wink:\"></p>",69 "post_number": 2,70 "post_type": 1,71 "posts_count": 2,72 "updated_at": "2020-05-14T03:19:19.095Z",73 "reply_count": 0,74 "reply_to_post_number": null,75 "quote_count": 0,76 "incoming_link_count": 0,77 "reads": 7,78 "readers_count": 6,79 "score": 1.4,80 "yours": false,81 "topic_id": 80974,82 "topic_slug": "model-execution-issues-with-cudnn-rnn",83 "display_username": "",84 "primary_group_name": null,85 "flair_name": null,86 "flair_url": null,87 "flair_bg_color": null,88 "flair_color": null,89 "flair_group_id": null,90 "badges_granted": [],91 "version": 1,92 "can_edit": false,93 "can_delete": false,94 "can_recover": false,95 "can_see_hidden_post": false,96 "can_wiki": false,97 "read": true,98 "user_title": "",99 "bookmarked": false,100 "actions_summary": [],101 "moderator": true,102 "admin": true,103 "staff": true,104 "user_id": 3534,105 "hidden": false,106 "trust_level": 2,107 "deleted_at": null,108 "user_deleted": false,109 "edit_reason": null,110 "can_view_edit_history": true,111 "wiki": false,112 "post_url": "/t/model-execution-issues-with-cudnn-rnn/80974/2",113 "can_accept_answer": false,114 "can_unaccept_answer": false,115 "accepted_answer": false,116 "topic_accepted_answer": null117 }118 ],119 "stream": [120 192223,121 192534122 ]123 },124 "timeline_lookup": [125 [126 1,127 1992128 ],129 [130 2,131 1991132 ]133 ],134 "suggested_topics": [135 {136 "fancy_title": "Torch using two GPUs with NV link",137 "id": 212483,138 "title": "Torch using two GPUs with NV link",139 "slug": "torch-using-two-gpus-with-nv-link",140 "posts_count": 9,141 "reply_count": 7,142 "highest_post_number": 10,143 "image_url": null,144 "created_at": "2024-11-03T22:15:14.885Z",145 "last_posted_at": "2024-11-05T09:29:43.969Z",146 "bumped": true,147 "bumped_at": "2024-11-05T09:29:43.969Z",148 "archetype": "regular",149 "unseen": false,150 "pinned": false,151 "unpinned": null,152 "visible": true,153 "closed": false,154 "archived": false,155 "bookmarked": null,156 "liked": null,157 "tags_descriptions": {},158 "like_count": 0,159 "views": 785,160 "category_id": 8,161 "featured_link": null,162 "has_accepted_answer": false,163 "posters": [164 {165 "extras": "latest",166 "description": "Original Poster, Most Recent Poster",167 "user": {168 "id": 77701,169 "username": "MLangner",170 "name": "",171 "avatar_template": "/letter_avatar_proxy/v4/letter/m/34f0e0/{size}.png",172 "trust_level": 1173 }174 },175 {176 "extras": null,177 "description": "Frequent Poster",178 "user": {179 "id": 3534,180 "username": "ptrblck",181 "name": "",182 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",183 "admin": true,184 "moderator": true,185 "trust_level": 2186 }187 }188 ]189 },190 {191 "fancy_title": "Initial D_KL loss is high and going down really slow",192 "id": 217877,193 "title": "Initial D_KL loss is high and going down really slow",194 "slug": "initial-d-kl-loss-is-high-and-going-down-really-slow",195 "posts_count": 1,196 "reply_count": 0,197 "highest_post_number": 1,198 "image_url": null,199 "created_at": "2025-03-15T12:04:06.828Z",200 "last_posted_at": "2025-03-15T12:04:06.870Z",201 "bumped": true,202 "bumped_at": "2025-03-15T12:07:55.407Z",203 "archetype": "regular",204 "unseen": false,205 "pinned": false,206 "unpinned": null,207 "visible": true,208 "closed": false,209 "archived": false,210 "bookmarked": null,211 "liked": null,212 "tags_descriptions": {},213 "like_count": 0,214 "views": 32,215 "category_id": 8,216 "featured_link": null,217 "has_accepted_answer": false,218 "posters": [219 {220 "extras": "latest single",221 "description": "Original Poster, Most Recent Poster",222 "user": {223 "id": 83290,224 "username": "User_Name",225 "name": "User Name",226 "avatar_template": "/user_avatar/discuss.pytorch.org/user_name/{size}/76177_2.png",227 "trust_level": 0228 }229 }230 ]231 },232 {233 "fancy_title": "Feed a model with cumulative sum of sampled classified sequences",234 "id": 216055,235 "title": "Feed a model with cumulative sum of sampled classified sequences",236 "slug": "feed-a-model-with-cumulative-sum-of-sampled-classified-sequences",237 "posts_count": 1,238 "reply_count": 0,239 "highest_post_number": 1,240 "image_url": null,241 "created_at": "2025-01-30T15:22:32.244Z",242 "last_posted_at": "2025-01-30T15:22:32.285Z",243 "bumped": true,244 "bumped_at": "2025-01-30T15:22:32.285Z",245 "archetype": "regular",246 "unseen": false,247 "pinned": false,248 "unpinned": null,249 "visible": true,250 "closed": false,251 "archived": false,252 "bookmarked": null,253 "liked": null,254 "tags_descriptions": {},255 "like_count": 0,256 "views": 32,257 "category_id": 8,258 "featured_link": null,259 "has_accepted_answer": false,260 "posters": [261 {262 "extras": "latest single",263 "description": "Original Poster, Most Recent Poster",264 "user": {265 "id": 82401,266 "username": "Seam1",267 "name": "Seam",268 "avatar_template": "/user_avatar/discuss.pytorch.org/seam1/{size}/75384_2.png",269 "trust_level": 1270 }271 }272 ]273 },274 {275 "fancy_title": "LSTM for classification (fraud detection) over several lines of text",276 "id": 216350,277 "title": "LSTM for classification (fraud detection) over several lines of text",278 "slug": "lstm-for-classification-fraud-detection-over-several-lines-of-text",279 "posts_count": 1,280 "reply_count": 0,281 "highest_post_number": 1,282 "image_url": null,283 "created_at": "2025-02-07T10:42:24.333Z",284 "last_posted_at": "2025-02-07T10:42:24.380Z",285 "bumped": true,286 "bumped_at": "2025-02-07T11:24:44.201Z",287 "archetype": "regular",288 "unseen": false,289 "pinned": false,290 "unpinned": null,291 "visible": true,292 "closed": false,293 "archived": false,294 "bookmarked": null,295 "liked": null,296 "tags_descriptions": {},297 "like_count": 0,298 "views": 144,299 "category_id": 8,300 "featured_link": null,301 "has_accepted_answer": false,302 "posters": [303 {304 "extras": "latest single",305 "description": "Original Poster, Most Recent Poster",306 "user": {307 "id": 82539,308 "username": "Monkee_Motion",309 "name": "Monkee Motion",310 "avatar_template": "/user_avatar/discuss.pytorch.org/monkee_motion/{size}/75523_2.png",311 "trust_level": 0312 }313 }314 ]315 },316 {317 "fancy_title": "How can I successfully fine-tune a pruned LLM?",318 "id": 221382,319 "title": "How can I successfully fine-tune a pruned LLM?",320 "slug": "how-can-i-successfully-fine-tune-a-pruned-llm",321 "posts_count": 7,322 "reply_count": 4,323 "highest_post_number": 7,324 "image_url": null,325 "created_at": "2025-07-09T11:35:56.843Z",326 "last_posted_at": "2025-07-09T16:39:47.538Z",327 "bumped": true,328 "bumped_at": "2025-07-17T13:00:28.603Z",329 "archetype": "regular",330 "unseen": false,331 "pinned": false,332 "unpinned": null,333 "visible": true,334 "closed": false,335 "archived": false,336 "bookmarked": null,337 "liked": null,338 "tags_descriptions": {},339 "like_count": 3,340 "views": 142,341 "category_id": 8,342 "featured_link": null,343 "has_accepted_answer": false,344 "posters": [345 {346 "extras": null,347 "description": "Original Poster",348 "user": {349 "id": 83456,350 "username": "Sixte_Oriol_Llenas_S",351 "name": "Sixte Oriol Llenas Segura",352 "avatar_template": "/user_avatar/discuss.pytorch.org/sixte_oriol_llenas_s/{size}/76334_2.png",353 "trust_level": 1354 }355 },356 {357 "extras": null,358 "description": "Frequent Poster",359 "user": {360 "id": 84935,361 "username": "paulk",362 "name": "",363 "avatar_template": "/letter_avatar_proxy/v4/letter/p/67e7ee/{size}.png",364 "trust_level": 2365 }366 },367 {368 "extras": "latest",369 "description": "Most Recent Poster",370 "user": {371 "id": 41458,372 "username": "J_Johnson",373 "name": "J Johnson",374 "avatar_template": "/user_avatar/discuss.pytorch.org/j_johnson/{size}/55494_2.png",375 "trust_level": 2376 }377 }378 ]379 }380 ],381 "tags_descriptions": {},382 "fancy_title": "Model execution issues with cuDNN RNN",383 "id": 80974,384 "title": "Model execution issues with cuDNN RNN",385 "posts_count": 2,386 "created_at": "2020-05-13T06:40:27.411Z",387 "views": 411,388 "reply_count": 0,389 "like_count": 0,390 "last_posted_at": "2020-05-14T03:19:19.095Z",391 "visible": true,392 "closed": false,393 "archived": false,394 "has_summary": false,395 "archetype": "regular",396 "slug": "model-execution-issues-with-cudnn-rnn",397 "category_id": 8,398 "word_count": 1409,399 "deleted_at": null,400 "user_id": 31513,401 "featured_link": null,402 "pinned_globally": false,403 "pinned_at": null,404 "pinned_until": null,405 "image_url": null,406 "slow_mode_seconds": 0,407 "draft": null,408 "draft_key": "topic_80974",409 "draft_sequence": null,410 "unpinned": null,411 "pinned": false,412 "current_post_number": 1,413 "highest_post_number": 2,414 "deleted_by": null,415 "actions_summary": [416 {417 "id": 4,418 "count": 0,419 "hidden": false,420 "can_act": false421 },422 {423 "id": 8,424 "count": 0,425 "hidden": false,426 "can_act": false427 },428 {429 "id": 10,430 "count": 0,431 "hidden": false,432 "can_act": false433 },434 {435 "id": 7,436 "count": 0,437 "hidden": false,438 "can_act": false439 }440 ],441 "chunk_size": 20,442 "bookmarked": false,443 "topic_timer": null,444 "message_bus_last_id": 0,445 "participant_count": 2,446 "show_read_indicator": false,447 "thumbnails": null,448 "slow_mode_enabled_until": null,449 "can_vote": false,450 "vote_count": 0,451 "user_voted": false,452 "discourse_zendesk_plugin_zendesk_id": null,453 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",454 "details": {455 "can_edit": false,456 "notification_level": 1,457 "participants": [458 {459 "id": 3534,460 "username": "ptrblck",461 "name": "",462 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",463 "post_count": 1,464 "primary_group_name": null,465 "flair_name": null,466 "flair_url": null,467 "flair_color": null,468 "flair_bg_color": null,469 "flair_group_id": null,470 "admin": true,471 "moderator": true,472 "trust_level": 2473 },474 {475 "id": 31513,476 "username": "mizho",477 "name": "Michael Zhou",478 "avatar_template": "/user_avatar/discuss.pytorch.org/mizho/{size}/13049_2.png",479 "post_count": 1,480 "primary_group_name": null,481 "flair_name": null,482 "flair_url": null,483 "flair_color": null,484 "flair_bg_color": null,485 "flair_group_id": null,486 "trust_level": 1487 }488 ],489 "created_by": {490 "id": 31513,491 "username": "mizho",492 "name": "Michael Zhou",493 "avatar_template": "/user_avatar/discuss.pytorch.org/mizho/{size}/13049_2.png"494 },495 "last_poster": {496 "id": 3534,497 "username": "ptrblck",498 "name": "",499 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png"500 }501 },502 "bookmarks": []503 },504 {505 "post_stream": {506 "posts": [507 {508 "id": 192533,509 "name": "",510 "username": "ngravindra",511 "avatar_template": "/letter_avatar_proxy/v4/letter/n/858c86/{size}.png",512 "created_at": "2020-05-14T03:18:09.329Z",513 "cooked": "<p>I want to create a wide array from long data.</p>\n<p>I have some edge indices, <code>edge_index=tensor([[0,0,0,1,1,1...], [0,2,3,1,2,3,..]])</code> for node 0 connected to node 0,2,3</p>\n<p>associated with three edge features, <code>edge_attr=tensor([[0.1,0.2,0.3], [0.4,0.5,0.6],...])</code></p>\n<p>such that edge_index.shape = 2 x n_edges, and edge_attr.shape= n_edges x 3 (using pytorch geometric’s data object).</p>\n<p>I am trying to re-shape the edge attributes to have shape [n_nodes, max_n_edges, 3] so each node, e.g., node 0, has its 3 connections 1st edge feature in one row, and its next edge feature concatenated in the third dimension.</p>\n<p>I can think of a pandas pivot way of doing this but the duplicates in the node indices (node 0 has 3 connections) poses somewhat of a problem and is slow… Is there a torch way of doing this?</p>",514 "post_number": 1,515 "post_type": 1,516 "posts_count": 1,517 "updated_at": "2020-05-14T03:18:09.329Z",518 "reply_count": 0,519 "reply_to_post_number": null,520 "quote_count": 0,521 "incoming_link_count": 34,522 "reads": 3,523 "readers_count": 2,524 "score": 170.6,525 "yours": false,526 "topic_id": 81128,527 "topic_slug": "long-to-wide-tensor-according-to-indices",528 "display_username": "",529 "primary_group_name": null,530 "flair_name": null,531 "flair_url": null,532 "flair_bg_color": null,533 "flair_color": null,534 "flair_group_id": null,535 "badges_granted": [],536 "version": 1,537 "can_edit": false,538 "can_delete": false,539 "can_recover": false,540 "can_see_hidden_post": false,541 "can_wiki": false,542 "read": true,543 "user_title": null,544 "bookmarked": false,545 "actions_summary": [],546 "moderator": false,547 "admin": false,548 "staff": false,549 "user_id": 31575,550 "hidden": false,551 "trust_level": 1,552 "deleted_at": null,553 "user_deleted": false,554 "edit_reason": null,555 "can_view_edit_history": true,556 "wiki": false,557 "post_url": "/t/long-to-wide-tensor-according-to-indices/81128/1",558 "can_accept_answer": false,559 "can_unaccept_answer": false,560 "accepted_answer": false,561 "topic_accepted_answer": null,562 "can_vote": false563 }564 ],565 "stream": [566 192533567 ]568 },569 "timeline_lookup": [570 [571 1,572 1991573 ]574 ],575 "suggested_topics": [576 {577 "fancy_title": "Loss.backward() causing unexpected errors",578 "id": 215842,579 "title": "Loss.backward() causing unexpected errors",580 "slug": "loss-backward-causing-unexpected-errors",581 "posts_count": 1,582 "reply_count": 0,583 "highest_post_number": 1,584 "image_url": null,585 "created_at": "2025-01-24T23:46:01.100Z",586 "last_posted_at": "2025-01-24T23:46:01.144Z",587 "bumped": true,588 "bumped_at": "2025-01-25T12:28:36.478Z",589 "archetype": "regular",590 "unseen": false,591 "pinned": false,592 "unpinned": null,593 "visible": true,594 "closed": false,595 "archived": false,596 "bookmarked": null,597 "liked": null,598 "tags_descriptions": {},599 "like_count": 0,600 "views": 36,601 "category_id": 1,602 "featured_link": null,603 "has_accepted_answer": false,604 "posters": [605 {606 "extras": "latest single",607 "description": "Original Poster, Most Recent Poster",608 "user": {609 "id": 82300,610 "username": "gabe_j",611 "name": null,612 "avatar_template": "/letter_avatar_proxy/v4/letter/g/bb73d2/{size}.png",613 "trust_level": 1614 }615 }616 ]617 },618 {619 "fancy_title": "Non blocking copy from CPU to GPU",620 "id": 213522,621 "title": "Non blocking copy from CPU to GPU",622 "slug": "non-blocking-copy-from-cpu-to-gpu",623 "posts_count": 2,624 "reply_count": 0,625 "highest_post_number": 2,626 "image_url": null,627 "created_at": "2024-11-27T13:17:50.447Z",628 "last_posted_at": "2024-11-28T09:42:41.192Z",629 "bumped": true,630 "bumped_at": "2024-11-28T09:42:41.192Z",631 "archetype": "regular",632 "unseen": false,633 "pinned": false,634 "unpinned": null,635 "visible": true,636 "closed": false,637 "archived": false,638 "bookmarked": null,639 "liked": null,640 "tags_descriptions": {},641 "like_count": 0,642 "views": 162,643 "category_id": 1,644 "featured_link": null,645 "has_accepted_answer": false,646 "posters": [647 {648 "extras": "latest single",649 "description": "Original Poster, Most Recent Poster",650 "user": {651 "id": 81162,652 "username": "shira",653 "name": "shira",654 "avatar_template": "/user_avatar/discuss.pytorch.org/shira/{size}/74225_2.png",655 "trust_level": 1656 }657 }658 ]659 },660 {661 "fancy_title": "Torch.nn.functional.embedding with output argument?",662 "id": 212632,663 "title": "Torch.nn.functional.embedding with output argument?",664 "slug": "torch-nn-functional-embedding-with-output-argument",665 "posts_count": 1,666 "reply_count": 0,667 "highest_post_number": 1,668 "image_url": null,669 "created_at": "2024-11-06T20:34:34.616Z",670 "last_posted_at": "2024-11-06T20:34:34.723Z",671 "bumped": true,672 "bumped_at": "2024-11-06T20:34:34.723Z",673 "archetype": "regular",674 "unseen": false,675 "pinned": false,676 "unpinned": null,677 "visible": true,678 "closed": false,679 "archived": false,680 "bookmarked": null,681 "liked": null,682 "tags_descriptions": {},683 "like_count": 0,684 "views": 59,685 "category_id": 1,686 "featured_link": null,687 "has_accepted_answer": false,688 "posters": [689 {690 "extras": "latest single",691 "description": "Original Poster, Most Recent Poster",692 "user": {693 "id": 9970,694 "username": "youkaichao1",695 "name": "",696 "avatar_template": "/letter_avatar_proxy/v4/letter/y/f0a364/{size}.png",697 "trust_level": 2698 }699 }700 ]701 },702 {703 "fancy_title": "Performance issue fitting multiple models with CPU in parallel",704 "id": 215318,705 "title": "Performance issue fitting multiple models with CPU in parallel",706 "slug": "performance-issue-fitting-multiple-models-with-cpu-in-parallel",707 "posts_count": 2,708 "reply_count": 0,709 "highest_post_number": 2,710 "image_url": null,711 "created_at": "2025-01-13T07:25:58.523Z",712 "last_posted_at": "2025-01-16T14:03:59.333Z",713 "bumped": true,714 "bumped_at": "2025-01-16T14:03:59.333Z",715 "archetype": "regular",716 "unseen": false,717 "pinned": false,718 "unpinned": null,719 "visible": true,720 "closed": false,721 "archived": false,722 "bookmarked": null,723 "liked": null,724 "tags_descriptions": {},725 "like_count": 0,726 "views": 57,727 "category_id": 1,728 "featured_link": null,729 "has_accepted_answer": false,730 "posters": [731 {732 "extras": "latest single",733 "description": "Original Poster, Most Recent Poster",734 "user": {735 "id": 82048,736 "username": "carusyte",737 "name": "",738 "avatar_template": "/letter_avatar_proxy/v4/letter/c/5fc32e/{size}.png",739 "trust_level": 0740 }741 }742 ]743 },744 {745 "fancy_title": "5070Ti+Ubuntu 20.04.6+cuda?",746 "id": 217901,747 "title": "5070Ti+Ubuntu 20.04.6+cuda?",748 "slug": "5070ti-ubuntu-20-04-6-cuda",749 "posts_count": 3,750 "reply_count": 1,751 "highest_post_number": 3,752 "image_url": null,753 "created_at": "2025-03-16T03:22:30.520Z",754 "last_posted_at": "2025-03-19T10:15:03.040Z",755 "bumped": true,756 "bumped_at": "2025-03-19T10:15:03.040Z",757 "archetype": "regular",758 "unseen": false,759 "pinned": false,760 "unpinned": null,761 "visible": true,762 "closed": false,763 "archived": false,764 "bookmarked": null,765 "liked": null,766 "tags_descriptions": {},767 "like_count": 0,768 "views": 273,769 "category_id": 1,770 "featured_link": null,771 "has_accepted_answer": false,772 "posters": [773 {774 "extras": "latest",775 "description": "Original Poster, Most Recent Poster",776 "user": {777 "id": 83301,778 "username": "riva_lei",779 "name": "riva lei",780 "avatar_template": "/user_avatar/discuss.pytorch.org/riva_lei/{size}/76187_2.png",781 "trust_level": 1782 }783 },784 {785 "extras": null,786 "description": "Frequent Poster",787 "user": {788 "id": 3534,789 "username": "ptrblck",790 "name": "",791 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",792 "admin": true,793 "moderator": true,794 "trust_level": 2795 }796 }797 ]798 }799 ],800 "tags_descriptions": {},801 "fancy_title": "Long to wide tensor according to indices",802 "id": 81128,803 "title": "Long to wide tensor according to indices",804 "posts_count": 1,805 "created_at": "2020-05-14T03:18:09.268Z",806 "views": 373,807 "reply_count": 0,808 "like_count": 0,809 "last_posted_at": "2020-05-14T03:18:09.329Z",810 "visible": true,811 "closed": false,812 "archived": false,813 "has_summary": false,814 "archetype": "regular",815 "slug": "long-to-wide-tensor-according-to-indices",816 "category_id": 1,817 "word_count": 157,818 "deleted_at": null,819 "user_id": 31575,820 "featured_link": null,821 "pinned_globally": false,822 "pinned_at": null,823 "pinned_until": null,824 "image_url": null,825 "slow_mode_seconds": 0,826 "draft": null,827 "draft_key": "topic_81128",828 "draft_sequence": null,829 "unpinned": null,830 "pinned": false,831 "current_post_number": 1,832 "highest_post_number": 1,833 "deleted_by": null,834 "actions_summary": [835 {836 "id": 4,837 "count": 0,838 "hidden": false,839 "can_act": false840 },841 {842 "id": 8,843 "count": 0,844 "hidden": false,845 "can_act": false846 },847 {848 "id": 10,849 "count": 0,850 "hidden": false,851 "can_act": false852 },853 {854 "id": 7,855 "count": 0,856 "hidden": false,857 "can_act": false858 }859 ],860 "chunk_size": 20,861 "bookmarked": false,862 "topic_timer": null,863 "message_bus_last_id": 0,864 "participant_count": 1,865 "show_read_indicator": false,866 "thumbnails": null,867 "slow_mode_enabled_until": null,868 "can_vote": false,869 "vote_count": 0,870 "user_voted": false,871 "discourse_zendesk_plugin_zendesk_id": null,872 "discourse_zendesk_plugin_zendesk_url": "https://your-url.zendesk.com/agent/tickets/",873 "details": {874 "can_edit": false,875 "notification_level": 1,876 "participants": [877 {878 "id": 31575,879 "username": "ngravindra",880 "name": "",881 "avatar_template": "/letter_avatar_proxy/v4/letter/n/858c86/{size}.png",882 "post_count": 1,883 "primary_group_name": null,884 "flair_name": null,885 "flair_url": null,886 "flair_color": null,887 "flair_bg_color": null,888 "flair_group_id": null,889 "trust_level": 1890 }891 ],892 "created_by": {893 "id": 31575,894 "username": "ngravindra",895 "name": "",896 "avatar_template": "/letter_avatar_proxy/v4/letter/n/858c86/{size}.png"897 },898 "last_poster": {899 "id": 31575,900 "username": "ngravindra",901 "name": "",902 "avatar_template": "/letter_avatar_proxy/v4/letter/n/858c86/{size}.png"903 }904 },905 "bookmarks": []906 },907 {908 "post_stream": {909 "posts": [910 {911 "id": 192224,912 "name": "seungwan seo",913 "username": "lepoeme20",914 "avatar_template": "/user_avatar/discuss.pytorch.org/lepoeme20/{size}/17561_2.png",915 "created_at": "2020-05-13T06:41:23.094Z",916 "cooked": "<p>Hi,</p>\n<p>Recently, I studied graph and learned that torch has torch geometric.</p>\n<p>I’m using CUDA 10.2 on ubuntu 18.04 and I tried installed using conda, but it failed.</p>\n<p>How can I use torch geometric with CUDA10.2?</p>",917 "post_number": 1,918 "post_type": 1,919 "posts_count": 2,920 "updated_at": "2020-05-13T06:41:23.094Z",921 "reply_count": 0,922 "reply_to_post_number": null,923 "quote_count": 0,924 "incoming_link_count": 220,925 "reads": 6,926 "readers_count": 5,927 "score": 1101.2,928 "yours": false,929 "topic_id": 80975,930 "topic_slug": "torch-geometric-on-cuda-10-2",931 "display_username": "seungwan seo",932 "primary_group_name": null,933 "flair_name": null,934 "flair_url": null,935 "flair_bg_color": null,936 "flair_color": null,937 "flair_group_id": null,938 "badges_granted": [],939 "version": 1,940 "can_edit": false,941 "can_delete": false,942 "can_recover": false,943 "can_see_hidden_post": false,944 "can_wiki": false,945 "read": true,946 "user_title": "",947 "bookmarked": false,948 "actions_summary": [],949 "moderator": false,950 "admin": false,951 "staff": false,952 "user_id": 24219,953 "hidden": false,954 "trust_level": 1,955 "deleted_at": null,956 "user_deleted": false,957 "edit_reason": null,958 "can_view_edit_history": true,959 "wiki": false,960 "post_url": "/t/torch-geometric-on-cuda-10-2/80975/1",961 "can_accept_answer": false,962 "can_unaccept_answer": false,963 "accepted_answer": false,964 "topic_accepted_answer": true,965 "can_vote": false966 },967 {968 "id": 192532,969 "name": "",970 "username": "ptrblck",971 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",972 "created_at": "2020-05-14T03:17:21.997Z",973 "cooked": "<p>Based on <a href=\"https://github.com/rusty1s/pytorch_geometric#pytorch-150\">this information</a>, their binaries should be available with CUDA10.2 for Windows and Linux.<br>\nWhat kind of error did you get?</p>",974 "post_number": 2,975 "post_type": 1,976 "posts_count": 2,977 "updated_at": "2020-05-14T12:15:23.775Z",978 "reply_count": 0,979 "reply_to_post_number": null,980 "quote_count": 0,981 "incoming_link_count": 4,982 "reads": 6,983 "readers_count": 5,984 "score": 36.2,985 "yours": false,986 "topic_id": 80975,987 "topic_slug": "torch-geometric-on-cuda-10-2",988 "display_username": "",989 "primary_group_name": null,990 "flair_name": null,991 "flair_url": null,992 "flair_bg_color": null,993 "flair_color": null,994 "flair_group_id": null,995 "badges_granted": [],996 "version": 1,997 "can_edit": false,998 "can_delete": false,999 "can_recover": false,1000 "can_see_hidden_post": false,1001 "can_wiki": false,1002 "link_counts": [1003 {1004 "url": "https://github.com/rusty1s/pytorch_geometric#pytorch-150",1005 "internal": false,1006 "reflection": false,1007 "title": "GitHub - rusty1s/pytorch_geometric: Geometric Deep Learning Extension Library for PyTorch",1008 "clicks": 491009 }1010 ],1011 "read": true,1012 "user_title": "",1013 "bookmarked": false,1014 "actions_summary": [1015 {1016 "id": 2,1017 "count": 11018 }1019 ],1020 "moderator": true,1021 "admin": true,1022 "staff": true,1023 "user_id": 3534,1024 "hidden": false,1025 "trust_level": 2,1026 "deleted_at": null,1027 "user_deleted": false,1028 "edit_reason": null,1029 "can_view_edit_history": true,1030 "wiki": false,1031 "post_url": "/t/torch-geometric-on-cuda-10-2/80975/2",1032 "can_accept_answer": false,1033 "can_unaccept_answer": false,1034 "accepted_answer": true,1035 "topic_accepted_answer": true1036 }1037 ],1038 "stream": [1039 192224,1040 1925321041 ]1042 },1043 "timeline_lookup": [1044 [1045 1,1046 19921047 ],1048 [1049 2,1050 19911051 ]1052 ],1053 "suggested_topics": [1054 {1055 "fancy_title": "cuda/Indexing.cu:1422: indexSelectLargeIndex: block: [381,0,0], thread: [66,0,0] Assertion `srcIndex < srcSelectDimSize` failed.RuntimeError: CUDA error: device-side assert triggered Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. Cou",1056 "id": 219154,1057 "title": "cuda/Indexing.cu:1422: indexSelectLargeIndex: block: [381,0,0], thread: [66,0,0] Assertion `srcIndex < srcSelectDimSize` failed.RuntimeError: CUDA error: device-side assert triggered Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. Cou",1058 "slug": "cuda-indexing-cu-indexselectlargeindex-block-381-0-0-thread-66-0-0-assertion-srcindex-srcselectdimsize-failed-runtimeerror-cuda-error-device-side-assert-triggered-compile-with-torch-use-cuda-dsa-to-enable-device-side-assertions-cou",1059 "posts_count": 2,1060 "reply_count": 0,1061 "highest_post_number": 2,1062 "image_url": null,1063 "created_at": "2025-04-16T12:04:51.018Z",1064 "last_posted_at": "2025-04-16T13:32:37.569Z",1065 "bumped": true,1066 "bumped_at": "2025-04-16T13:32:37.569Z",1067 "archetype": "regular",1068 "unseen": false,1069 "pinned": false,1070 "unpinned": null,1071 "visible": true,1072 "closed": false,1073 "archived": false,1074 "bookmarked": null,1075 "liked": null,1076 "unicode_title": "cuda/Indexing.cu:1422: indexSelectLargeIndex: block: [381,0,0], thread: [66,0,0] Assertion `srcIndex < srcSelectDimSize` failed.RuntimeError: CUDA error: device-side assert triggered Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions. Cou",1077 "tags_descriptions": {},1078 "like_count": 0,1079 "views": 121,1080 "category_id": 1,1081 "featured_link": null,1082 "has_accepted_answer": false,1083 "posters": [1084 {1085 "extras": null,1086 "description": "Original Poster",1087 "user": {1088 "id": 83839,1089 "username": "chenGlin",1090 "name": "chenGlin",1091 "avatar_template": "/user_avatar/discuss.pytorch.org/chenglin/{size}/76649_2.png",1092 "trust_level": 01093 }1094 },1095 {1096 "extras": "latest",1097 "description": "Most Recent Poster",1098 "user": {1099 "id": 3534,1100 "username": "ptrblck",1101 "name": "",1102 "avatar_template": "/user_avatar/discuss.pytorch.org/ptrblck/{size}/1823_2.png",1103 "admin": true,1104 "moderator": true,1105 "trust_level": 21106 }1107 }1108 ]1109 },1110 {1111 "fancy_title": "Weight normalization of lazily initialized modules",1112 "id": 216641,1113 "title": "Weight normalization of lazily initialized modules",1114 "slug": "weight-normalization-of-lazily-initialized-modules",1115 "posts_count": 1,1116 "reply_count": 0,1117 "highest_post_number": 1,1118 "image_url": null,1119 "created_at": "2025-02-13T19:03:54.446Z",1120 "last_posted_at": "2025-02-13T19:03:54.488Z",1121 "bumped": true,1122 "bumped_at": "2025-02-13T19:03:54.488Z",1123 "archetype": "regular",1124 "unseen": false,1125 "pinned": false,1126 "unpinned": null,1127 "visible": true,1128 "closed": false,1129 "archived": false,1130 "bookmarked": null,1131 "liked": null,1132 "tags_descriptions": {},1133 "like_count": 0,1134 "views": 71,1135 "category_id": 1,1136 "featured_link": null,1137 "has_accepted_answer": false,1138 "posters": [1139 {1140 "extras": "latest single",1141 "description": "Original Poster, Most Recent Poster",1142 "user": {1143 "id": 82669,1144 "username": "davidkwho",1145 "name": "David Ho",1146 "avatar_template": "/user_avatar/discuss.pytorch.org/davidkwho/{size}/75647_2.png",1147 "trust_level": 11148 }1149 }1150 ]1151 },1152 {1153 "fancy_title": "Questions Regarding the Direction of Half Padding for Odd-Sized Inputs in PyTorch Conv2D and ConvTransposed2D",1154 "id": 214237,1155 "title": "Questions Regarding the Direction of Half Padding for Odd-Sized Inputs in PyTorch Conv2D and ConvTransposed2D",1156 "slug": "questions-regarding-the-direction-of-half-padding-for-odd-sized-inputs-in-pytorch-conv2d-and-convtransposed2d",1157 "posts_count": 2,1158 "reply_count": 0,1159 "highest_post_number": 2,1160 "image_url": "https://discuss.pytorch.org/uploads/default/optimized/3X/b/b/bbbaac4cedae04ffa60904b9c583ad66ddfa697c_2_1024x449.png",1161 "created_at": "2024-12-15T07:28:33.352Z",1162 "last_posted_at": "2024-12-17T13:51:47.078Z",1163 "bumped": true,1164 "bumped_at": "2024-12-17T13:51:47.078Z",1165 "archetype": "regular",1166 "unseen": false,1167 "pinned": false,1168 "unpinned": null,1169 "visible": true,1170 "closed": false,1171 "archived": false,1172 "bookmarked": null,1173 "liked": null,1174 "tags_descriptions": {},1175 "like_count": 0,1176 "views": 91,1177 "category_id": 1,1178 "featured_link": null,1179 "has_accepted_answer": false,1180 "posters": [1181 {1182 "extras": "latest single",1183 "description": "Original Poster, Most Recent Poster",1184 "user": {1185 "id": 81511,1186 "username": "mx34kryce",1187 "name": "",1188 "avatar_template": "/user_avatar/discuss.pytorch.org/mx34kryce/{size}/74521_2.png",1189 "trust_level": 11190 }1191 }1192 ]1193 },1194 {1195 "fancy_title": "There is not libgomp-a34b3233.so.1 in torch/lib when I build the torch based on the V2.4.0 with USE_CUDA=0",1196 "id": 218511,1197 "title": "There is not libgomp-a34b3233.so.1 in torch/lib when I build the torch based on the V2.4.0 with USE_CUDA=0",1198 "slug": "there-is-not-libgomp-a34b3233-so-1-in-torch-lib-when-i-build-the-torch-based-on-the-v2-4-0-with-use-cuda-0",1199 "posts_count": 1,1200 "reply_count": 0,